Optimization and design of an aircraft's morphing wing­tip demonstrator for drag 

reduction at low speeds, Part II ­ Experimental validation using Infra­Red 

transition measurement from Wind Tunnel tests

Koreanschi, A, Sugar­Gabor, O, Acotto, J, Brianchon, G, Portier, G, Botez, RM, Mamou, M and Mebarki, Y

http://dx.doi.org/10.1016/j.cja.2016.12.018

Title Optimization and design of an aircraft's morphing wing­tip demonstrator for drag reduction at low speeds, Part II ­ Experimental validation using Infra­Red transition measurement from Wind Tunnel tests

Authors Koreanschi, A, Sugar­Gabor, O, Acotto, J, Brianchon, G, Portier, G, Botez, RM, Mamou, M and Mebarki, Y

Type Article

URL This version is available at: http://usir.salford.ac.uk/id/eprint/41503/

Published Date 2017

USIR is a digital collection of the research output of the University of Salford. Where copyright permits, full text material held in the repository is made freely available online and can be read, downloaded and copied for non­commercial private study or research purposes. Please check the manuscript for any further copyright restrictions.

For more information, including our policy and submission procedure, pleasecontact the Repository Team at: usir@salford.ac.uk.

Chinese Journal of Aeronautics, (2017), 30(1): 164–174

Chinese Society of Aeronautics and Astronautics& Beihang University

Chinese Journal of Aeronautics

cja@buaa.edu.cnwww.sciencedirect.com

Optimization and design of an aircraft’s morphing

wing-tip demonstrator for drag reduction at low

speeds, Part II - Experimental validation using

Infra-Red transition measurement from Wind

Tunnel tests

* Corresponding author.

E-mail addresses: koreanschiandreea@yahoo.com (A. Koreanschi), ruxandra.botez@etsmtl.ca (R.M. Botez).

Peer review under responsibility of Editorial Committee of CJA.

Production and hosting by Elsevier

http://dx.doi.org/10.1016/j.cja.2016.12.0181000-9361 � 2017 Chinese Society of Aeronautics and Astronautics. Production and hosting by Elsevier Ltd.This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Andreea Koreanschi a, Oliviu Sugar Gabor a, Joran Acotto a, Guillaume Brianchon a,

Gregoire Portiera, Ruxandra Mihaela Botez

a,*, Mahmoud Mamoub,

Youssef Mebarkib

aLARCASE Laboratory of Applied Research in Active Control, Avionics and Aeroservoelasticity, Ecole de TechnologieSuperieure, Montreal H3C1K3, CanadabAerodynamics Laboratory, NRC Aerospace, National Research Council Canada, Ottawa K1A0R6, Canada

Received 8 March 2016; revised 13 June 2016; accepted 21 June 2016Available online 3 January 2017

KEYWORDS

Drag reduction;

Infra-red tests;

Morphing wing;

Optimization;

Wind tunnel tests

Abstract In the present paper, an ‘in-house’ genetic algorithm was numerically and experimentally

validated. The genetic algorithm was applied to an optimization problem for improving the aero-

dynamic performances of an aircraft wing tip through upper surface morphing. The optimization

was performed for 16 flight cases expressed in terms of various combinations of speeds, angles of

attack and aileron deflections. The displacements resulted from the optimization were used during

the wind tunnel tests of the wing tip demonstrator for the actuators control to change the upper

surface shape of the wing. The results of the optimization of the flow behavior for the airfoil mor-

phing upper-surface problem were validated with wind tunnel experimental transition results

obtained with infra-red Thermography on the wing-tip demonstrator. The validation proved that

the 2D numerical optimization using the ‘in-house’ genetic algorithm was an appropriate tool in

improving various aspects of a wing’s aerodynamic performances.� 2017 Chinese Society of Aeronautics and Astronautics. Production and hosting by Elsevier Ltd. This is

an open access article under the CCBY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Optimization and design of an aircraft’s morphing wing-tip demonstrator 165

1. Introduction

Nowadays, applications of optimization algorithms can befound in almost all industrial and academic research venues,

such as optimization electric circuitry,1 stock market predic-tions,2 image quality problems,3 software implementationproblems,4 to optimization of aircraft structures, aerodynam-

ics or flight trajectories, etc.In the aerospace field, many research projects and collabo-

rations include the successful implementation of the more tra-ditional metaheuristic optimization algorithms such as genetic

algorithm,5 bee colony algorithm,6 artificial neural net-works,7,8 or ant colonies optimization in their research fornew optimized flight trajectories, for new optimized wing

shapes or improved control.One such collaboration took place between the teams from

the Laboratory of Applied Research in Active Control, Avion-

ics and Aeroservoelasticity (LARCASE) laboratory and CMCelectronics-esterline on the Green Aviation Research Develop-ment Network (GARDN) project, which was funded by the

Green Aviation Research Development Business Led Networkin its second round.9,10 The main objective of the collaborationwas to optimize the vertical and horizontal paths of the aircraftwithin the flight management system by taking into account

the required time of arrival, the wind grids and meteorologicalconditions. The main motivation of the project was to reduceoverall carbon emissions and flight costs.

Morphing also consists in changing the structure or appear-ance of an aircraft during flight by modifying the wingsweep,11 span,12 chord13 or camber,14,15 by the high lift

devices16,17 or the fuselage, for small aircraft and forunmanned aerial vehicle (UAV).18,19

Applications of optimization techniques for UAVs were

described by Gamboa et al.20 who designed an UAV wing cap-able of independent span and chord changes, using a telescopicspar and a rib system. The numerical analysis demonstrated adrag reduction of up to 23% when compared to its non-

morphing base geometry. Falcao et al.21 designed and testeda morphing winglet for a military UAV and achieved impor-tant performance improvements by changing the winglet cant

and toe angles. Other research on UAV wing morphing wasdone by Sugar et al.,22,23 where the upper-surface of the wingwas optimized on a segment between its leading edge and 55%

of the chord, and also explored morphing of the full wing’sgeometry. Hu and Yu24 developed a multi-disciplinary opti-mization for improving aerodynamic, stealth and structuralperformances of an unmanned aerial combat vehicle. Li et al.25

developed a methodology for aerodynamic optimization aimedat demonstrating the performances of a blended wing bodytransport, while Xie et al.26 studied the effects of static aeroe-

lastic phenomena on very flexible wings.Few projects concentrate on the effect of the morphing

technologies on the aerodynamic performances of the wing;

the majority concentrate mostly on aerodynamic and struc-tural interactions for the purpose of demonstrating theincreased safety against undesired aeroelastic phenomena such

as flutter.27–29

A recent experiment, where the aerodynamic performancesof active morphing wings were studied, was the CRIAQ 7.1project, in which collaboration took place between aerospace

industrial teams from Bombardier Aerospace and Thales

Canada, and academic partners from the Ecole de TechnologieSuperieure (ETS) and Ecole Polytechnique of Montreal, andthe Canadian National Research Council (CNRC) team. The

purpose of the project was to demonstrate the capabilities ofmorphing wings in a wind tunnel for developing the flow tran-sition from laminar to turbulent.30,31 Morphing was achieved

by replacing the upper surface of the wing, spanned between7% and 70% of the wing chord, by a flexible carbon-Kevlarcomposite skin. The skin morphing was achieved using two

shape memory alloy (SMA) actuation lines with the aim toobtain an optimized shape for each flight condition tested inthe wind tunnel.32 The optimization was done using a geneticalgorithm method coupled with the aerodynamic solver XFoil.

The wind tunnel tests had proven that the concept of uppersurface morphing was viable, controllable, and provided tangi-ble results by confirming the delay of the transition from lam-

inar to turbulent flow, which induced a substantial reductionin the drag coefficient.33 Proportional integrated derivative(PID)34 and neuro-fuzzy controllers35 were tested to prove

the controllability of the flexible skin shape and the morphingmechanisms towards the transition delay. It appeared that thecontrollers demonstrated an excellent performance in both

open36 and closed loops.37

The research presented in this present paper was donewithin the framework of the international CRIAQ MDO505Morphing Wing Project, which was a continuation of the pre-

vious research project CRIAQ 7.1, and aimed at a higher tech-nical readiness level by considering a real wing internalstructure and a certifiable electric control system and con-

trollers. The participants in this project were ETS, Ecole Poly-tehnique and University of Naples ‘Federico II’ as academiaresearch partners, the CNRC and the Italian Aerospace

Research Center (CIRA) as research center partners and Bom-bardier Aeronautique, Thales Canada and Alenia Aermacchias industrial partners.

The objectives of the project were to design, manufactureand control a wing demonstrator based on an aircraft wing-tip equipped with both a conventional and adaptive aileron.The novelty of the CRIAQ MDO 505 project consisted in its

multidisciplinary approach, where structure, aerodynamics,control and experimental design were combined to designand manufacture an active morphing wing demonstrator and

test it under subsonic wind tunnel conditions.Part I of this paper established the design and optimization

of a wing-tip demonstrator airfoil using an ‘in-house’ genetic

algorithm coupled with the XFoil aerodynamic 2D solver thatused the eN method for the numerical determination of thetransition point.38,39 The algorithm was described in detail,and its results were compared with the results obtained by

other optimization methods, namely the bee colony methodand the gradient method. Also, another experimental valida-tion of the genetic algorithm was performed for the ATR 42

wing airfoil in Koreanschi et al.40 Validation of the optimiza-tion technique and numerical results were achieved throughexperimental data obtained through wind tunnel tests of a

wing model demonstrator. The optimization concentrated onthe improvement of the upper-surface behavior of the flowby manipulating the position of transition from fully laminar

to fully turbulent flow. The optimization was carried at the air-foil level and in practice, was applied to a full scale wing tipwith aircraft-look-alike internal structure. The validation wasdone through comparison of the numerical and experimental

166 A. Koreanschi et al.

results for a specific region on the wing, where Kulite sensorswere installed for pressure measurements.

2. Wing tip demonstrator with conventional aileron

The full-scale morphing wing model was an optimized struc-ture with a 1.5 m span and 1.5 m root chord, a taper ratio of

0.72 and leading and trailing edges sweep angle of 8�. The wingbox and its internal structure (spars, ribs, and lower skin) weremanufactured from aluminum alloy material, while the adap-

tive upper surface was positioned between 20% and 65% ofthe wing chord. The adaptive upper surface skin was specifi-cally designed and optimized to meet industrial partner’s

requirements. The adaptive skin was manufactured using car-bon fiber composite materials.41

The deformation of the skin shape, driven by actuators

placed inside the wing box structure, was a function of theflight condition (defined in terms of Mach number, Reynoldsnumber and angle of attack). These actuators were specificallydesigned and manufactured to meet in-flight and wind tunnel

test requirements. Four electrical actuators were installed ontwo actuation lines; two actuators were installed on each line,were placed at 37% and 75% of the wing span, and were fixed

to the ribs and to the composite skin. Each actuator has theability to operate independently from the others. On each actu-ation line, the actuators were positioned at 32% and 48% of

the local wing chord.The aileron’s hinge was located at 72% of the chord. Two

ailerons type were designed and manufactured. One aileronwas structurally rigid, while the other one represented a new

morphing aileron concept. Both ailerons were designed to beattached to the same hinge axis of the wing box, and both wereable to undergo a controlled deflection between �7� and +7�.Fig. 1 presents a sketch of the morphing wing model conceptas it was mounted and tested at the NRC subsonic windtunnel.

Fig. 1 CRIAQ MDO 505 morphing wing concept.

3. Wind tunnel description and infra-red data acquisition

The wind tunnel tests were performed at the 2 m � 3 m atmo-spheric closed circuit subsonic wind tunnel of the CNRC. This

atmospheric wind tunnel can operate at a maximum Machnumber of 0.33.

The upper surface flexible skin was equipped with 32 high

precision Kulite piezoelectric-type transducers42 for pressuremeasurement on the flexible skin that were further processedto determine the laminar-to-turbulent transition location.These sensors were installed in two staggered lines (with 16

Kulite sensors on each line), situated respectively at 0.600 mand 0.625 m from the wing root section. In addition to theKulite piezoelectric sensors, at the same two spanwise stations,

60 static pressure taps were installed (30 taps on each line) onthe wing leading edge, lower surface and aileron, thus provid-ing complete experimental pressure distribution around the

wing cross section at 40% of the wing span. The pressure sen-sors were installed in a staggered fashion to minimize the inter-ference between sensors.

The experimental measurements also included the use of awake rake pressure acquisition system for the purpose of mea-suring the wing profile drag at different span-wise positions,and also the use of a wind tunnel balance for measuring the

aerodynamic forces and moments. Fig. 2 presents the MDO505 morphing wing model installed in the tunnel test section,viewed from both the leading edge (LE) (Fig. 2(a)) and the

trailing edge (TE) (Fig. 2(b)).Infra-red (IR) thermography camera visualizations were

performed for capturing the transition region over the entire

wing model surface. The wing leading edge, its upper surfaceflexible skin and the aileron interface were coated with highemissivity black paint to improve the quality of the IR pho-

tographs. The span-wise stations, where the two pressure sen-sors lines were installed, were not painted, in order to notinfluence the pressure reading quality. A Jenoptik Variocamcamera,43 with a resolution of 640 by 480 pixels, was used to

measure the surface temperatures. This camera was equippedwith 60� lens in order to capture the flow transition on theentire upper surface of the wing.

The IR thermography visualization allowed the identifica-tion of the transition region between laminar and turbulentregimes, based on the analysis of the model surface tempera-

ture. Examples of infra-red photography results are given inSection 5. The turbulent flow regime increases the convectiveheat transfer between the model and the flow with respect tothe laminar boundary layer. As a result, a flow temperature

change, introduced by the wind tunnel heat exchanger system,

Fig. 2 MDO 505 wing model setup in wind tunnel test section.

Optimization and design of an aircraft’s morphing wing-tip demonstrator 167

will cause different temperature changes over the model,depending on the behavior of the boundary layer.

4. Optimization algorithm

The genetic algorithm was applied to the problem of airfoilupper-surface morphing. The problem objective was the search

of the optimum shapes for an airfoil through local thicknessmodifications with the aim to improve the upper surface flowand thus the aerodynamic performances of the wing’s airfoil.

The local wing thickness modification was obtainedthrough four actuations points, as described in the previoussection. The shape of the flexible upper-surface was obtained

Table 1 Morphing problem variable values for MDO 505

wing demonstrator airfoil.

Variable Value

Morphing surface start point (%c) 20

Morphing surface end point (%c) 65

No. of actuators/chord 2

LE actuator (%c) 32

TE actuator (%c) 48

Maximum displacement (mm) 3.5

Type of displacement Vertical in both directions

Requirements for actuators Dactuators < 6 mm

Fig. 3 Diagram of ‘in-ho

by an optimized combination of the four vertical displace-ments, representing the local ‘pushing and pulling’ actions offour electric actuators installed inside the wing box. The verti-

cal displacements resulted from the genetic optimization of thewing’s airfoil.

For the theoretical thin airfoil provided by Bombardier,

considered under the name CRIAQ MDO 505 wing demon-strator airfoil, the optimization and design approach was moreconservative in nature, as many structural requirements and

constraints were taken into account when performing theoptimization.

Table 1 presents the morphing surface limits, number andposition of actuators on each rib as well as the maximum dis-

placements, %c means the percentage of the chord.The problem of airfoil upper-surface morphing for

improvement of the aerodynamic behavior of wings is not a

problem with a single solution. More often than not, as itwas presented in Section 1 of this paper, there is an optimumregion where several possible solutions coexist and any of them

can be considered as the final solution to the problem.A full description of the methodology used for the opti-

mization algorithm and its numerical results was provided in

Section 1 of this paper. Fig. 3 presents the workflow diagramof the algorithm that was used for the optimization.

Table 2 presents the parameters used for the optimizationof the 16 cases tested during the wind tunnel tests of the wing

demonstrator.

use’ genetic algorithm.

Table 2 Input blocks and parameters for MDO 505 demonstrator airfoil.

Input block Parameter Value Observation

Optimization No. of individuals 40

No. of generations 20

Probability of mutation 1% % of total population

Amplitude of mutation 2% % of the maximum displacement value

Optimization objective The objective is given through weights associated with aerodynamic

characteristics, such as lift and drag coefficients and flow transition

Geometry Airfoil coordinates

Chord of the airfoil (m) 1.332

Morphing surface start

point

20% % of chord

Morphing surface end point 65% % of chord

No. of actuators 2 can accept up to 4

LE actuator 32% % of chord

TE actuator 48% % of chord

Maximum displacement of

the actuators (mm)

3.5

Type of displacement Both positive (push) and negative (pull) actions are allowed

Spline

reconstruction

Number of splines 8

Atmosphere

data

Density (kg/m3) 1.22

Dynamic viscosity (Pa�s) 1.82 � 10�5

Temperature (K) 293

Altitude (m) 0

Flight data No. of cases 16

Speed Range of Mach speeds

Angle of attack Range of angles

Aileron deflection Range of angles

Fig. 4 Example of IR results for Case 3 from Table 3 (un-

morphed wing demonstrator shown without aileron).

168 A. Koreanschi et al.

5. Optimization simulation vs experimental results

In this section, the optimization of the MDO 505 wing airfoil is

presented. The optimization was performed using the parame-ters provided in Section 4, Table 2. The optimization results,provided as actuator displacements in mm, were used by the

control team to perform the upper-surface morphing of thewing-tip demonstrator during the wind tunnel tests.

The results were presented as numerical transition pointsfor the wing section, and as experimental transition regions

extracted from Infrared Thermography for the same wing sec-tion where the pressure sensors were installed (Figs. 4 and 5).

The two sets of results, numerical and experimental, were

firstly compared to assess the agreement between numericaland experimental values, and secondly to assess the optimiza-tion success during experimental tests and compare it to the

numerical optimization expectation.The optimization was run for two main objectives: transi-

tion delay towards the trailing edge (Eq. (1)), which meanspossible drag coefficient reduction, and transition advance-

ment towards the leading edge (Eq. (2)), which could stabilizethe boundary layer at high speeds or high angles of attack andaileron deflections.

Ff ¼ 100� UpTr morphed �UpTr original

UpTr original

� �ð1Þ

Ff ¼ 100� UpTr morphed �UpTr original

UpTr original

� �2

ð2Þ

where Ff represents the fitness function and UpTr representsthe airfoil’s upper surface transition position.

Table 3 presents the 16 cases studied and the numericalresults obtained with the genetic algorithm optimization for

both objective functions.The experimental tests were done at the CNRC subsonic

wind tunnel located in Ottawa/Ontario. The wind tunnel and

Fig. 5 Example of IR results for Case 7 from Table 3 (morphed

wing demonstrator shown without aileron).

Optimization and design of an aircraft’s morphing wing-tip demonstrator 169

the MDO 505 wing demonstrator used during the experimentswere described in the above Section 2 of the present paper.

The experimental transition location results were obtainedwith IR thermography; the results for the section of intereston the wing were extracted using MATLAB software; the IR

system was described in Section 3. The IR data post-processing steps consisted of: correction of the lens distortions,of the perspective view and projection onto the physical geom-

etry. The detection of the transition region was fully auto-mated by looking at the local temperature gradients on thewing surface. The final outputs of the data analysis were: the

transition region (delimited by white dotted lines on theimages), the mean transition front spanning the whole wingspan, and the mean transition at the Kulite pressure sensorsstation to compare with the CFD simulations. Figs. 4 and 5

present examples of IR results for three of the cases fromTable 3.

Table 3 Optimization cases and results for wing tip demonstrator.

Case No. Ma Angle of attack (�) Aileron deflection (�) Type of op

1 0.15 0.68 0 Delay tran

2 0.15 1.50 0 Delay tran

3 0.15 2.10 0 Delay tran

4 0.15 �2.39 2 Delay tran

5 0.15 1.93 �2 Delay tran

6 0.20 1.88 4 Delay tran

7 0.20 3.03 4 Delay tran

8 0.20 3.45 �4 Delay tran

9 0.15 �0.33 5 Advance t

10 0.15 �0.95 �2 Advance t

11 0.25 �2.99 1 Advance t

12 0.25 �2.26 3 Advance t

13 0.15 �2.30 2 Advance t

14 0.15 �1.64 3 Advance t

15 0.15 �3.22 �2 Advance t

16 0.25 �1.52 5 Advance t

The black dashed lines in Figs. 4 and 5 correspond to thesection of the wing demonstrator where the Kulite pressuresensors were installed, and also, represent the section chord

for which the optimization was performed. The optimizationwas done for the section where the first line of actuators wasinstalled, then it was linearly extrapolated for the second line

of actuators, which is close to the tip of the wing demonstrator.The experimental transition was presented as a ‘region’ and

the numerical transition point obtained with XFoil’s eN

method was matched to this region. If the numerical transitionpoint was inside the experimental transition region, then it wasconsidered that the numerical and experimental results were ingood agreement. If the numerical transition was outside the

experimental transition region, then an error was calculatedbetween the numerical value and the closest boundary value.If the calculated error was less than 6%, the error was consid-

ered as acceptable.44

Fig. 6 presents an example where the numerical transitionmatched the experimental transition region and an example

where the numerical transition did not match.As shown in Fig. 6, the numerical transition point was

found to be situated inside the experimental transition region

boundaries for Case 5, and in this case, a good agreementbetween numerical and experimental data existed, while inCase 6, the numerical transition was situated with 6% of thechord outside the lowest boundary of the experimental transi-

tion region, and it was viewed as having an acceptable errorbetween numerical and experimental transition.

5.1. Comparison between numerical and experimental transitiondata

Figs. 7 and 8 show the comparison that was made between the

numerically determined transition point and the experimentaltransition region from Infrared readings for the un-morphed,and for the morphed wing demonstrator. This comparison

was done to show the agreement between the numerical andthe experimental transition data.

timization Transition (%c) Improvement (%c)

Original airfoil Optimized airfoil

sition 53.62 54.47 0.85

sition 48.35 53.85 5.5

sition 46.09 52.41 6.32

sition 63.71 66.19 2.48

sition 43.34 52.97 9.63

sition 41.91 53.82 11.91

sition 33.44 50.62 17.18

sition 30.35 41.30 10.95

ransition 74.90 43.05 �31.85

ransition 60.01 50.92 �9.09

ransition 60.09 44.92 �15.17

ransition 59.46 45.05 �14.41

ransition 65.58 44.01 �21.57

ransition 67.43 43.48 �23.95

ransition 64.83 44.27 �20.56

ransition 64.52 41.77 �22.75

Fig. 6 Comparison between Case 5 when numerical transition

has matched experimental region and Case 6 when numerical

transition was found outside experimental region.

Fig. 7 Comparison between numerical transition point and

experimental transition region for the first 8 cases.

Fig. 8 Comparison between numerical transition point and

experimental transition region for the second set of Cases 9–16.

170 A. Koreanschi et al.

It was possible to successfully compare the numericalresults obtained for the wing’s airfoil to the experimental tran-sition results extracted for a specific section corresponding to

Kulite sensors localization from the global experimental resultsof the entire wing demonstrator.

In Figs. 7 and 8, the presented results show that with the

exception of 3 un-morphed wing cases (Cases 6, 7 and 9), thenumerical transition was situated inside the experimental tran-sition boundaries.

Tables 4 and 5 present the errors found for the 16 casesdescribed in Table 3; Table 4 presents the errors for the un-morphed wing demonstrator transition results and Table 5for the morphed wing demonstrator transition results.

The error was calculated as the difference between thenumeric transition value and the closest experimental transi-tion region boundary:

Error ¼ Transitionnum � Transitionexp ð3ÞWhen the error is 0 the numerical transition was situated

inside the experimental transition region.

5.2. Evaluation of experimental transition optimization

This section presents the behavior of the upper-surface morph-

ing during experimental testing on the MDO 505 wing demon-strator. In Fig. 9, the experimental un-morphed and morphedwing section transition regions were overlapped for a better

view of the effects of the upper-surface morphing on the lengthand position of the transition region in the studied section.

The experimental transition region is characterized by an

upper and a lower boundary. The lower boundary of the tran-sition region represents the point where the flow starts its tran-sition from fully laminar flow towards turbulent, while the

upper boundary of the transition region represents the locationat which the flow can be considered as being fully turbulent.Therefore, the optimization of the transition region refers tomodifications in the desired direction of the upper and lower

boundaries, depending on the optimization objective to beaccomplished.

As such, two parameters were calculated: s, which repre-

sents the difference between the morphed and un-morphedtransition region (TR) upper boundary values and describeshow much the onset of the fully turbulent flow was modified,

s ¼ TRMorphed;UB � TRUnmorphed;UB ð4Þwhere UB means upper-boundary, and k, which represents thedifference between the morphed and un-morphed TR lower

boundary values and describes with how much the boundaryof the fully laminar flow was modified.

k ¼ TRMorphed;LB � TRUnmorphed;LB ð5Þwhere LB means lower-boundary.

Fig. 9(a) shows the comparison between the un-morphedand morphed wing transition regions for the objective of flow

transition delay from fully laminar to fully turbulent. It couldbe observed from the above figure that the onset of the fullyturbulent flow was delayed for 7 cases out of 8, with the max-

imum delay being achieved for Case 7 with 7.65%c. The end ofthe laminar flow was also delayed in 6 cases, with the maxi-mum delay being again for Case 7 with 5.65%c. For Case 1,

the transition region of the morphed wing was extended in

Table 4 Transition intervals and values for numerical and experimental cases and error between the results (un-morphed wing).

Case No. Xfoil un-morphed (%c) Experimental un-morphed (%c) Error (%c)

Upper boundary Lower boundary Average

1 53.62 52.57 48.57 50.57 1.09

2 48.35 49.91 45.91 47.91 0

3 46.09 51.26 45.26 48.26 0

4 63.71 66.30 62.30 64.30 0

5 43.34 48.73 42.73 45.73 0

6 41.91 50.35 48.35 49.35 �6.44

7 33.44 43.69 41.69 42.69 �8.25

8 30.35 40.20 36.20 38.20 �5.85

9 74.90 66.22 64.22 65.22 8.68

10 60.01 57.70 47.70 52.70 2.31

11 60.09 55.35 51.35 53.35 4.74

12 59.46 55.28 51.28 53.28 4.18

13 65.58 65.83 61.83 63.83 0

14 67.43 65.79 63.79 64.79 1.64

15 64.83 65.73 65.73 65.73 0

16 64.52 55.80 53.80 54.80 8.72

Table 5 Transition intervals and values for numerical and experimental cases and error between the results (morphed wing).

Case No. Xfoil morphed (%c) Experimental morphed (%c) Error (%c)

Upper boundary Lower boundary Average

1 54.47 53.54 45.54 49.54 0.93

2 53.85 53.67 47.67 50.67 0.18

3 52.41 53.44 47.44 50.44 0

4 66.19 66.95 62.95 64.95 0

5 52.97 47.63 41.63 44.63 5.34

6 53.82 53.68 49.68 51.68 0.14

7 50.62 51.34 47.34 49.34 0

8 41.30 42.39 38.39 40.39 0

9 43.05 48.55 46.55 47.55 �3.50

10 50.92 52.13 46.13 49.13 0

11 44.92 47.49 43.49 45.49 0

12 45.05 47.73 43.73 45.73 0

13 44.01 48.41 46.41 47.41 0

14 43.48 48.95 44.95 46.95 �1.47

15 44.27 47.09 45.09 46.09 �0.82

16 41.77 45.91 41.91 43.91 �0.14

Optimization and design of an aircraft’s morphing wing-tip demonstrator 171

comparison with the original wing, while for Case 4 the differ-ence between the two regions was almost negligible. Case 5 was

the one case where the transition optimization was not success-ful, but the difference between the two transition regions wasalso very small.

Table 6 presents the values for the two parametersdescribed in the first part of the section, s and k, for the caseswhere the optimization was aimed at delaying the transition

from laminar towards turbulence of the upper-surface flow.Fig. 9(b) shows the comparison between the un-morphed

and morphed wing transition regions for the objective ofadvancing transition towards the leading edge.

From Fig. 9(b), it appeared that the onset of the fully tur-bulent flow was advanced towards the leading edge for allcases, with the maximum advancement being achieved for

Case 15 with 18.64%c. The end of the laminar flow was alsoadvanced towards the leading edge in all cases, with the max-imum advancement being again for case 15 of 20.64%c. For

Cases 10 and 13 the length of the transition region was reducedthrough the morphing of the upper surface, while for Cases

14–16 the length of the transition region was a little bitextended; all the other cases had an unchanged length of thetransition region.

Table 7 presents the values for the two parametersdescribed in the first part of the section, s and k, for the caseswhere the optimization was aimed at advancing the transition

on the wing upper-surface.Fig. 10 displays a comparison between the numerical tran-

sition optimization prediction and the resulted experimentaloptimization. Fig. 10(a) shows the comparison between the

numerical optimization prediction based on XFoil resultsand the s and k results with the objective to delay transition,while Fig. 10(b) presents the comparison between the numeri-

cal prediction and the s and k results with the objective ofadvancing transition. The two figures assess the differences

Fig. 9 Comparison between experimental un-morphed and

morphed transition regions.

172 A. Koreanschi et al.

between the numerical optimization predictions and the exper-imental results.

From Fig. 10, it could be observed that for most of the

cases the numerical optimization had overestimated the transi-tion delay or advancement, with some cases where the differ-ence is almost double. For Cases 1–4, 10 and 15 the

numerical prediction was close to the transition obtainedexperimentally during the wind tunnel tests.

The overestimation of the transition optimization cannot

be imputed to a single aspect or point in a single directionwhere an error could be found; as the designed and manufac-

Table 6 Parameters k and s describing effects of morphing wing o

Case No. Ma Angle of attack (�)

1 0.15 0.68

2 0.15 1.50

3 0.15 2.10

4 0.15 �2.39

5 0.15 1.93

6 0.20 1.88

7 0.20 3.03

8 0.20 3.45

Table 7 Parameters k and s describing effects of morphing wing

objective.

Case No. Ma Angle of attack (�)

9 0.15 �0.33

10 0.15 �0.95

11 0.25 �2.99

12 0.25 �2.26

13 0.15 �2.30

14 0.15 �1.64

15 0.15 �3.22

16 0.25 �1.52

tured MDO 505 wing demonstrator was the result of a multi-disciplinary project, where many aerospace disciplinesinteracted, any variation of any of the multiple variables per-

taining to structure, aerodynamics, control, integration orexperiment could have affected the outcome of the results.Nonetheless, despite the existing differences between the

numerical predictions and the experimental results, the opti-mization of the MDO 505 wing through morphing of theupper surface by using actuator displacements resulted from

a numerical optimization with an ‘in-house’ Genetic Algorithmcoupled with a bi-dimensional aerodynamic solver using the eN

method was considered as successful.

6. Conclusions

In this paper, an ‘in-house’ genetic algorithm was applied to

the problem of optimizing the shape of the upper surface ofan airfoil by using actuator displacements. In the first part ofthe paper it was shown that the genetic algorithm used forthe optimization of the wing tip demonstrator airfoil gave very

good results in comparison with two other optimization meth-ods and it always reached the global optimum region. It wasshown that the algorithm was robust and that it converged

towards the optimum area in less than 10 iterations or gener-ations, while other 10 generations were used to ensure the sta-bility of the solution and that this solution was found in the

global optimum area.Finally, the genetic algorithm was used to optimize the air-

foil shape for 16 cases, with the aim to satisfy two objectives:delay of the transition towards the trailing edge of the airfoil,

and advancement of the flow transition towards the leadingedge. The displacements resulted from the optimization wereused for the upper surface morphing controller during wind

tunnel testing on the MDO 505 wing demonstrator and

n flow behavior for transition delay objective.

Aileron deflection (�) s (%c) k (%c)

0 0.97 �3.03

0 3.76 1.76

0 2.19 2.19

2 0.66 0.66

�2 �1.10 �1.10

4 3.33 1.33

4 7.65 5.65

�4 2.19 2.19

on flow behavior, for transition advance towards leading edge

Aileron deflection (�) s (%c) k (%c)

5 17.67 17.67

�2 5.57 1.57

1 7.86 7.86

3 7.55 7.55

2 17.42 15.42

3 16.84 18.84

�2 18.64 20.64

5 9.89 11.89

Fig. 10 Comparison of numerical optimization transition and

experimental resulted optimization.

Optimization and design of an aircraft’s morphing wing-tip demonstrator 173

comparisons were conducted between the experimental transi-

tion regions of the morphed and un-morphed wing – section byusing Infrared Thermography. For the success of this opti-mization, two new parameters were introduced, s and k, todescribe the behavior of the flow when it passed from fullylaminar to fully turbulent. Both objectives were successfullyattained for most of the cases using the displacements provided

by the numerical optimization. Maximum delays of the transi-tion region were up to 7.6% of the chord and for the forwarddisplacement of the transition region were of up to 20% of thechord.

The experimental optimization results were then comparedwith the numerical simulation results, it was found that thenumerical optimization was overestimated due to a multitude

of factors starting with the numerical solver, and ending withthe multidisciplinary aspect of the project that introduced ahigh number of variables that could affect the numerical opti-

mization. Nonetheless, the numerical optimization was animportant tool for preliminary estimation and evaluation ofthe morphing possibilities and the GA presented in this paper

and could be successfully used for performing optimization ofthe wing’s upper-surface morphing problem. Also it would beinteresting to compare its results to those that could beobtained with more recent optimization methods such as those

based on mathematical behavior.

Acknowledgments

We would like to thank to Bombardier Aerospace, ThalesCanada, The Consortium in Research and Aerospace in

Canada (CRIAQ), and to the Natural Sciences and Engineer-ing Research Council of Canada (NSERC) for their financialsupport. Special thanks are dues to our collaborators and lead-

ers in this project: Mr. Patrick Germain and Mr. FassiKafyeke from Bombardier Aerospace, Mr. Eric Laurendeaufrom Ecole Polytechnique, Mr. Philippe Molaret from Thales

Canada, and Mr. Erik Sherwood and his team from DFS-

NRC for the wing model design and fabrication.

References

1. Zhang H, Ye D. An artificial bee colony algorithm approach for

routing in VLSI. Advances in swarm intelligence. Berlin Heidelberg:

Springer; 2012. p. 334–41.

2. Majhi R, Panda G, Majhi B, Sahoo G. Efficient prediction of

stock market indices using adaptive bacterial foraging optimiza-

tion (ABFO) and BFO based techniques. Expert Syst Appl 2009;36

(6):10097–104.

3. Cui SY, Wang ZH, Tsai PW, Chang CC, Yue S. Single bitmap

block truncation coding of color images using cat swarm optimiza-

tion. Recent advances in information hiding and applications. Berlin

Heidelberg: Springer; 2013. p. 119–38.

4. Bacanin N. Implementation and performance of an object-

oriented software system for cuckoo search algorithm. Int J Math

Comput Simul 2012;6(1):185–93.

5. Sugar OG, Koreanschi A, Botez RM. Low-speed aerodynamic

characteristics improvement of ATR 42 airfoil using a morphing

wing approach. IIECON 2012-38th annual conference on IEEE

industrial electronics society. Piscataway (NJ): IEEE Press; 2012.

p. 5451–6.

6. Gabor SO, Simon A, Koreanschi A, Botez RM. Aerodynamic

performance improvement of the UAS-S4 Ehecatl morphing

airfoil using novel optimization techniques. Proc Inst Mech Eng,

Part G: J Aerosp Eng 2016;230:1164–80.

7. Mosbah AB, Salinas MF, Botez RM, Dao TM. New methodology

for wind tunnel calibration using neural networks-EGD approach.

SAE Int J Aerosp 2013;6(2):761–6.

8. Mosbah AB, Botez RM, Dao TM. New methodology for

calculating flight parameters with neural network–EGD metho-

dAIAA modeling and simulation technologies (MST) confer-

ence. Reston: AIAA; 2013. p. 19–22.

9. Patron RF, Botez RM, Labour D. New altitude optimisation

algorithm for the flight management system CMA-9000 improve-

ment on the A310 and L-1011 aircraft. Roy Aeronaut Soc

2013;117:787–805.

10. Patron F, Salvador R, Kessaci A, Botez RM. Horizontal flight

trajectories optimisation for commercial aircraft through a flight

management system. Aeronaut J 2014;118(1210):1499–518.

11. Joo JJ, Sanders B, Johnson T, Frecker MI. Optimal actuator

location within a morphing wing scissor mechanism configuration.

Proceeding of SPIE 2006.

12. Neal DA, Good MG, Johnston CO, Robertshaw HH, Mason

DJ, Inman DJ. Design and wind-tunnel analysis of a fully adaptive

aircraft configurationProceedings of AIAA/ASME/ASCE/AHS/

ASC SDM. Reston: AIAA; 2004.

13. Reed Jr JL, Hemmelgarn CD, Pelley BM, Havens E. Adaptive

wing structures. Proceeding of SPIE 2005;132–42.

14. Poonsong P. Design and analysis of a multi-section variable

camber wing [dissertation]. Maryland: University of Maryland;

2004.

15. Monner HP, Hanselka H, Breitbach EJ. Development and design

of flexible fowler flaps for an adaptive wing. 5th annual interna-

tional symposium on smart structures and materials. 1998. p. 60–70.

16. Pecora R, Barbarino S, Lecce L, Russo S. Design and functional

test of a morphing high-lift device for a regional aircraft. J Intell

Mater Syst Struct 2011;22(10):1005–23.

17. Pecora R, Magnifico M, Amoroso F, Monaco E. Trade-off flutter

analysis of a morphing wing trailing edge. 6th ECCOMAS

conference on smart structures and materials. 2013.

18. Sugar OG, Koreanschi A, Botez RM. Optimization of an

unmanned aerial system’ wing using a flexible skin morphing

wing. Report No.: 2013-01-2095. Pennsylvania: SAE Interna-

tional; 2013.

174 A. Koreanschi et al.

19. Sugar OG, Simon A, Koreanschi A, Botez RM. Improving the

UAS-S4 Ehecal airfoil high angles-of-attack performance charac-

teristics using a morphing wing approach. Proc Inst Mech Eng,

Part G: J Aerosp Eng 2016;230(7):1164–80.

20. Gamboa P, Vale J, Lau PFJ, Suleman A. Optimization of a

morphing wing based on coupled aerodynamic and structural

constraints. AIAA J 2009;47(9):2087–104.

21. Falcao L, Gomes AA, Suleman A. Aero-structural design

optimization of a morphing wingtip. J Intell Mater Syst Struct

2011;22(10):1113–24.

22. Sugar OG, Koreanschi A, Botez RM. Numerical optimization

of the S4 Ehecatl UAS airfoil using a morphing wing approa-

chAIAA 32nd applied aerodynamics conference. Reston: AIAA;

2014.

23. Sugar OG, Simon A, Koreanschi A, Botez RM. Application of a

morphing wing technology on hydra technologies unmanned

aerial system UAS-S4ASME 2014 international mechanical engi-

neering congress and exposition. New York: ASME; 2014.

24. Hu TY, Yu XQ. Aerodynamic/stealthy/structural multidisci-

plinary design optimization of unmanned combat air vehicle.

Chin J Aeronaut 2009;22(4):380–6.

25. Li PF, Zhang B, Chen YC, Yuan CS, Lin Y. Aerodynamic design

methodology for blended wing body transport. Chin J Aeronaut

2012;25(4):508–16.

26. Xie CC, Wang L, Yang C, Liu Y. Static aeroelastic analysis of

very flexible wings based on non-planar vortex lattice method.

Chin J Aeronaut 2013;26(3):514–21.

27. Liauzun C. Aeroelastic response to gust using CFD tech-

niquesASME 2010 3rd joint US-European fluids engineering

summer meeting collocated with 8th international conference on

nanochannels, microchannels, and minichannels. New York: ASME;

2010. p. 269–76.

28. Pecora R, Amoroso F, Lecce L. Effectiveness of wing twist

morphing in roll control. J Aircraft 2012;49(6):1666–74.

29. Pecora R, Magnifico M, Amoroso F, Monaco E. Multi-paramet-

ric flutter analysis of a morphing wing trailing edge. Aeronaut J

2014;118(1207):1063–78.

30. Popov AV, Botez RM, Mamou M, Me´barki Y, Jahrhaus B,

Khalid M, et al. Drag reduction by improving laminar flows past

morphing configurations. AVT-168 NATO symposium on the

morphing vehicles. 2009.

31. Botez RM, Molaret P, Laurendeau E. Laminar flow control on a

research wing project presentation covering a three year period.

Canadian Aeronautics and Space Institute annual general meeting

2007.

32. Grigorie TL, Botez RM, Popov AV, Mamou M, Mebarki Y. A

hybrid fuzzy logic proportional-integral-derivative and conven-

tional on-off controller for morphing wing actuation using shape

memory alloy—Part 1: Morphing system mechanisms and con-

troller architecture design. Aeronaut J 2012;116(1179):433–49.

33. Sainmont C, Paraschivoiu I, Coutu D, Brailovski V, Laurendeau

M, Mamou M, et al. Boundary layer behaviour on a morphing

airfoil: simulation and wind tunnel tests. CASI AERO’09 confer-

ence aerodynamics symposium. 2009.

34. Grigorie TL, Botez RM, Popov AV. Design and experimental

validation of a control system for a morphing wingAIAA

atmospheric flight mechanics conference. Reston: AIAA; 2012.

35. Grigorie LT, Botez RM, Popov AV, Mamou M, Mebarki Y. A

new morphing mechanism for a wing using smart actuators

controlled by a self-tuning fuzzy logic controller. AIAA centennial

of naval aviation forum. Reston: AIAA; 2011.

36. Popov AV, Grigorie LT, Botez RM, Mamou M, Mebarki Y. Real

time morphing wing optimization validation using wind-tunnel

tests. J Aircraft 2010;47(4):1346–55.

37. Popov AV, Grigorie LT, Botez RM, Mamou M, Mebarki Y.

Closed-loop control validation of a morphing wing using wind

tunnel tests. J Aircraft 2010;47(4):1309–17.

38. Drela M, Youngren D. XFOIL version 6.96 [Internet], 2001.

Available from: http://web.mit.edu/aeroutil_v1.0/xfoil_doc.txt

[cited 2016 Mar 8].

39. Drela M. Implicit implementation of the full eN transition

criterion. Report No.: AIAA-2003-4066. Reston: AIAA; 2003.

40. Koreanschi A, Sugar O, Botez RM. Numerical and experimental

validation of a morphed wing geometry using price- Paıdoussis

wind tunnel testing. Aeronaut J 2016;120(1227):757–95.

41. Michaud F. Design and optimization of a composite skin for an

adaptive wing [dissertation]. Montreal: Ecole de Technology

Superieure; 2014.

42. Kulite Semiconductor Products [Internet]. Available from: http://

Kulite.Com/ [cited 2016 Mar 8].

43. Mebarki Y, Mamou M, Genest M. Infrared measurements of the

transition detection on the CRIAQ project morphing wing model.

Report No.: NRC LTR AL-2009-007, 2009.

44. Robitaille M, Mosahebi A, Laurendau E. Design of adaptive

transonic laminar airfoils using the c-Reht transition model. Aerosp

Sci Technol 2015;46:60–71.